Hacker Newsnew | past | comments | ask | show | jobs | submit | cpburns2009's commentslogin

If you have the hardware to run expensive models, is the cost of electricity much of a factor? According to Google, the average price in the Silicon Valley Area is $0.448 per kWh. An RTX 5090 costs about $4,000 and has a peak power consumption of 1000 W. Maxing out that GPU for a whole year would cost $3,925 at that rate. It's not particularly more expensive than that hardware itself.

At that point it'd be cheaper to get an expensive subscription to a cloud platform AI product. I understand the case for local LLMs but it seems silly to worry about pricing for cloud-based offerings but not worry about pricing for locally run models. Especially since running it locally can often be more expensive

Blue light filtering lenses come at a premium. You don't accidentally get them.

Let me explain how many times I went back-and-forth with the opticians about "is this coating/feature optional?"

My optician's office charges an extra $100 for blue light filtering. They at least make it clear it's optional but recommended for frequent screen use.

I wasn't paying for them, so it was very much accidental.

You should go back and demand they be replaced. Such a mistake isn't something you should tolerate.

I don't go to that optician anymore.

The list of mistakes were as follows:

1) it corrected my eye with a slight astigmatism, but over corrected my other eye, so the agregate was pretty much the same

2) the aformentioned blue filter, which is part of the anti-glare coating.

3) my non-astigmatic eye was incorrectly marked with a prescription

4) I don't actually need glasses because I can see to the bottom of the eye chart without them. Its just as I'm now older, my vision is not as good as they used to be.

5) these were designed for "close work" but actually don't really help me focus closer to me.

However, arguing that, as a non-proffesional with only a passing understanding of optics (non-biological) with a large multinational company doesn't seem like a good use of time.


Someone ran up to you and put them on your face?

Why would you want your face to submit a form? How could it even be selected by default? Is this an April Fools joke?

Say you want to have a custom button element. You cant give it a `type=submit` whilst you can set that on <input> and <button>

Also if you have a face that wraps a button of type submit. The submit doesn't propagate due to shadow dom


Does "face" have some sort of special meaning in this context? I thought it might be related an open web implementation of Face ID.

From the title of the issue: “Form-associated custom elements”. They are talking about the mechanism by which a custom element can have its value included when you submit a form. Here’s the spec.:

https://html.spec.whatwg.org/multipage/custom-elements.html#...

https://html.spec.whatwg.org/multipage/form-control-infrastr...

https://html.spec.whatwg.org/multipage/form-control-infrastr...


I realize that the GP is trolling, but he accidentally has a good point. It is useful to know what censorship goes into a model. Apparently Copilot censors words related to gender and refuses to autocomplete them. That'd be frustrating to work with. A Chinese model censoring terms for events they want to memory hole would have zero impact on anything I would ever work on.

You're looking for text-to-speech. Qwen actually has a model and library for this: Qwen3-TTS [1].

[1]: https://github.com/QwenLM/Qwen3-TTS


I benchmarked unsloth/Qwen3-Coder-Next-GGUF using the MXFP4_MOE (43.7 GB) quantization on my Ryzen AI Max+ 395 and I got ~30 tps. According to [1] and [2], the AI Max+ 395 is 2.4x faster than the AI 9 HX 370 (laptop edition). Taking all that into account, the AI 9 HX 370 should get ~13 tps on this model. Make of that what you will.

[1]: https://community.frame.work/t/ai-9-hx-370-vs-ai-max-395/736...

[2]: https://community.frame.work/t/tracking-will-the-ai-max-395-...


Thanks! I'm... unimpressed.

In my short testing on a different MoE model, it does not perform well. I tried running Kimi-K2-Thinking-GGUF with the smallest unsloth quantization (UD-TQ1_0, 247 GB), and it ran at 0.1 tps. According to its guide, you should expect 5 tps if the whole model can fit into RAM+VRAM, but if mmap has to be used, then expect less than 1 tps which matches my test. This was on a Ryzen AI Max+ 395 using ~100 GB VRAM.

Running a 247GB model reliably on 100GB VRAM total is a very impressive outcome no matter what the performance. That size of model is one where sensible people will recommend at least 4x the VRAM amount compared to what you were testing with - at that point, the total bandwidth to your storage becomes the bottleneck. Try running models that are just slightly bigger than the amount of VRAM you're using and these tricks become quite essential, for a significantly more manageable hit on performance.

That's NVME storage in your test?

Yes, a WD_BLACK 4TB SN850X NVMe.

You don't have to statically allocate the VRAM in the BIOS. It can be dynamically allocated. Jeff Geerling found you can reliably use up to 108 GB [1].

[1]: https://www.jeffgeerling.com/blog/2025/increasing-vram-alloc...


Generic WASP checking in. I flew regularly for several years until covid and I'd get screened all the time too (about 50% of the time).


Now I haven't touched C++ in probably 15 years but the definition of main() looks confused:

> auto main() -> int

Isn't that declaring the return type twice, once as auto and the other as int?


No. The auto there is doing some lifting so that you can declare the type afterwards. The return type is only defined once.

There is, however, a return type auto-deduction in recent standards iirc, which is especially useful for lambdas.

https://en.cppreference.com/w/cpp/language/auto.html

auto f() -> int; // OK: f returns int

auto g() { return 0.0; } // OK since C++14: g returns double

auto h(); // OK since C++14: h’s return type will be deduced when it is defined


What about

auto g() -> auto { return 0.0; }


0.0 is a double, so I would assume the return type of g is deduced to be double, if that is what you're asking.


I was more pointing out that the syntax was dumb for that particular example!


I really wish they had used func instead, it would have saved this confusion and allowed for “auto type deduction” to be a smaller more self contained feature


the standard c++ committee is extremely resistant to introducing new keywords such as "func", so as not to break reams of existing code.


Indeed. I am a frequent critic of the c++ committee’s direction and decisions. There’s no direction other than “new stuff” and that new stuff pretty much has to be in the library otherwise it will require changes that may break existing code. That’s fine.

But on the flip side, there’s a theme of ignoring the actual state of the world to achieve the theoretical goals of the proposal when it suits. Modules are a perfect example of this - when I started programming professionally modules were the solution to compile times and to symbol visibility. Now that they’re here they are neither. But we got modules on part. The version that was standardised refused to accept the existence of the toolchain and build tools that exist, and as such refused to place any constraints that may make implementation viable or easier.

St the same time we can’t standardise Pragma once because some compiler may treat network shares or symlinks differently.

There’s a clear indication that the committee don’t want to address this, epochs are a solution that has been rejected. It’s clear the only real plan is shove awkward functional features into libraries using operator overloads - just like we all gave out to QT for doing 30 years ago. But at least it’s standardised this time?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: